Per Stenstrom

Deadline approaching: fully funded positions for PhD students on two exciting topics! Radar Radiance Maps for Robust Perception and Navigation with Martin Magnusson and Henrik Andreasson. Working towards towards truly robust, long-term, all-weather operation for self-driving vehicles and robots; specifically using modern 3D data representations such as neural radiance fields and Gaussian splatting, the use of state-of-the-art high-resolution radar sensors -- and the combination thereof. Apply by August 19 at https://lnkd.in/dMHC9hrT Robotic Lab Automation for Handling Powdered and Granular Materials with Johannes A. Stork and Yasemin B. Going beyond handling of tubes and beakers, handling of powdered and granular chemical substances is essential for analytic and production processes. This line of research will include perceiving and modeling loose material using force-torque and vibratory sensing, planning and executing scooping actions including online sensor feedback. Apply by August 30 at https://lnkd.in/df5SiGuy Örebro University WASP – Wallenberg AI, Autonomous Systems and Software Program

29

PhD student Martin Alexandersson at the Division of Marine Technology at Chalmers University of Technology, and RISE Research Institutes of Sweden, has a new paper published in the Ocean Engineering journal: "System identification of a physics-informed ship model for better predictions in wind conditions". The paper proposes a new physics-informed ship manoeuvring model, where a deterministic semi-empirical rudder model has been added, to guide the identification towards a physically correct hydrodynamic model. This is an essential building block to distinguish the hydrodynamic modelling uncertainties from wind, waves, and currents – in real sea conditions – which is particularly important for ships with wind-assisted propulsion. In the physics-informed manoeuvring modelling framework, a systematical procedure is developed to establish various force/motion components within the manoeuvring system by inverse dynamics regression. https://lnkd.in/d4eC5MBm

29

“Theory is all you need.” Provocative thesis by Teppo Felin presented at our Research Frontiers seminar at Aalto TUTA - Aalto University, Department of Industrial Engineering and Management. He and his co-author Matthias Holweg argue that there’s a fundamental difference between how deep learning and generative AI work and how humans think. The latest generation of AI technologies is based on finding and using correlations in massive bodies of data and is essentially backward-looking, whereas humans are imaginative and forward-looking. In their view human intelligence is not about prediction but about problem solving and intervention in the world based on hypotheses and “theories”. Rather than seeing humans as input-output machines, humans (and other organisms as well) have theories that determine what data to consider (and what to discard or ignore) in the first place. Teppo uses several examples to illustrate their claim. Galileo was quite foolish to think the earth revolves around the sun, since all data and evidence pointed to the opposite. The Wright brothers were equally foolish in believing that heavier than air flight is possible, as most experts disagreed with them, and even though birds can fly, obviously birds that are too heavy can't. In hindsight we know they were right, though. They imagined novel theories. They consciously ignored most of the available and instead focused on very specific data that informed their theories and helped them test their hypotheses. For example, when it comes to making heavier than air flight possible, looking at birds' weight is misleading, but instead one should look at, for example, the shape of their wings. I think they are right in their claim, though of course not in the title of their paper (there's a few more things needed). Myself being a strategy scholar, their arguments resonate with me, as strategy is essentially about imagination and action in order to bring about a desired future. A great strategy builds on, but also breaks with the past, so it's more than recombination (which generative AI is good at). When it comes to AI, clearly generative AI is a massive breakthrough, and it's amazing what LLMs can do. But my intuition is that we are still a bit away from reconstructing anything close to human intelligence in machines. Or at least from machines substituting for the strategist. Teppo is also one of the "fathers" of the emerging theory-based view in strategy. I myself have also contributed to this literature, see the comments for a link to a recent paper. I also post a link to Teppo's paper there. #artificialintelligence #generativeai #humanintelligence #cognition #theory #strategy

62

4 kommentarer

New preprint out with Lei Du at Chalmers tekniska högskola and the Wallenberg Centre for Quantum Technology (WACQT): ”Unconventional and robust light-matter interactions based on the non-Hermitian skin effect” Here we study a series of unconventional light-matter interactions between quantum emitters and the prototypical Hatano–Nelson model as well as an extended lattice model dubbed the bosonic Kitaev chain. We focus on the robustness of the dynamics to various imperfections. We consider both small emitters, which interact with the lattice at single sites, and giant emitters coupling at multiple sites. The latter exhibit an exclusive amplification mechanism, which we find enables decoherence-free dynamics even with on-site dissipation.

29

A long wait but this paper is finally accepted to IEEE Transactions on Neural Networks and Learning Systems #TNNLS From Bi-Level to One-Level: A Framework for Structural Attacks to Graph Anomaly Detection https://lnkd.in/ghAxZdMp

8

In today's rapidly evolving technological landscape, staying ahead of the curve is more crucial than ever. That's why attending a conference on printed electronics can be a game-changer. Here are some compelling reasons to mark your calendar: 💡 Innovative Insights: Discover the latest advancements in printed electronics, from flexible displays to wearable tech. Gain first-hand knowledge from leading experts and innovators shaping the future of this dynamic field. 💲 Business Growth: Learn how printed electronics can drive efficiency, reduce costs, and open new revenue streams. Explore how integrating these technologies can transform your business and give you a competitive edge. 🍓 Sustainability: Printed electronics often involve materials and processes that are more sustainable compared to traditional electronics. Discover how this technology can contribute to your sustainability goals and corporate social responsibility initiatives. Don't miss out on this unique opportunity to expand your horizons and propel your business forward. Join us at the next printed electronics conference in the ❤️ of Norrköping and be part of the revolution! Register now: https://lnkd.in/d_7dWe3h

15